• Laser & Optoelectronics Progress
  • Vol. 57, Issue 22, 221507 (2020)
Jing Zhang, Zhihui Hao, and Jing Liu*
Author Affiliations
  • School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China
  • show less
    DOI: 10.3788/LOP57.221507 Cite this Article Set citation alerts
    Jing Zhang, Zhihui Hao, Jing Liu. Template-Updating Algorithm Based on Optical Flow Mapping in Object Tracking[J]. Laser & Optoelectronics Progress, 2020, 57(22): 221507 Copy Citation Text show less
    Schematic of proposed tracking algorithm framework
    Fig. 1. Schematic of proposed tracking algorithm framework
    Results of ablation experiments on OTB100 dataset. (a) Success rate; (b) precision
    Fig. 2. Results of ablation experiments on OTB100 dataset. (a) Success rate; (b) precision
    Subjective and objective (average precision) results of ablation experiments in Coke sequence
    Fig. 3. Subjective and objective (average precision) results of ablation experiments in Coke sequence
    Subjective and objective (average precision) results of ablation experiments in Dudek sequence
    Fig. 4. Subjective and objective (average precision) results of ablation experiments in Dudek sequence
    Subjective and objective (average precision) results of ablation experiments in Doll sequence
    Fig. 5. Subjective and objective (average precision) results of ablation experiments in Doll sequence
    Experimental results of different algorithms on OTB100 dataset. (a) Success rate; (b) precision
    Fig. 6. Experimental results of different algorithms on OTB100 dataset. (a) Success rate; (b) precision
    Comparison results of different algorithms when target moves rapidly on OTB100 dataset. (a) Success rate; (b) precision
    Fig. 7. Comparison results of different algorithms when target moves rapidly on OTB100 dataset. (a) Success rate; (b) precision
    Subjective and objective (average precision) results of each algorithm in Coke sequence
    Fig. 8. Subjective and objective (average precision) results of each algorithm in Coke sequence
    Subjective and objective (average precision) results of each algorithm in Dudek sequence
    Fig. 9. Subjective and objective (average precision) results of each algorithm in Dudek sequence
    Subjective and objective (average precision) results of each algorithm in Doll sequence
    Fig. 10. Subjective and objective (average precision) results of each algorithm in Doll sequence
    NetworkEAO↑Robustness↓Lost number↓Accuracy↑Speed/(frame·s-1)↑
    B (baseline)0.4640.19642.00.64232.3
    B+W (ours)0.4740.19141.00.64231.6
    B+W+C (ours+)0.4780.17237.00.64032.0
    Table 1. Results of ablation experiments on VOT2016 dataset
    NetworkEAO↑Robustness↓Accuracy↑
    SRDCF0.3380.240.51
    C-COT0.3310.240.51
    ECO-HC0.3220.300.54
    Siam-FC0.2350.460.53
    SiamFCRes220.3030.380.54
    SiamRPNRes220.3760.240.58
    TADT0.2990.310.55
    C-RPN0.3630.270.59
    SPM0.4340.210.62
    ours+0.4780.170.64
    Table 2. Experimental results of different algorithms on VOT2016 dataset
    Jing Zhang, Zhihui Hao, Jing Liu. Template-Updating Algorithm Based on Optical Flow Mapping in Object Tracking[J]. Laser & Optoelectronics Progress, 2020, 57(22): 221507
    Download Citation